The 22nd CCF National High Performance Computing Academic Conference (CCF HPC2026) Held in Shenzhen
The 22nd CCF National High Performance Computing Academic Conference (CCF HPC2026) was held on October 17-20 at the Shenzhen World Exhibition & Convention Center, co-located in the same venue and concurrently with the 2026 World Computing Power Expo, under the theme "Deep Integration of Supercomputing and Intelligence, Shenzhen's Rich Computing Power." The conference featured 11 invited talks, 8 corporate presentations, and 50 thematic forums, bringing together more than 20 Chinese and foreign academicians and over 500 experts and scholars. The total exhibition area exceeded 12,000 square meters, with 65 companies exhibiting. At the opening ceremony, the HPC+AI Supercomputing-Intelligence Collaborative Open Community and the High-Throughput Ethernet (ETH+) Alliance's full-stack domestic intelligent computing network were released.
Topics
Body · AI translation
Where the tide rises toward the sea in Pengcheng, computing power surges forth to launch a new journey. From October 17 to 20, the 22nd CCF National High Performance Computing Academic Conference (CCF HPC2026) officially opened at the Shenzhen World Exhibition & Convention Center. Held concurrently and in the same venue as the 2026 World Computing Expo (WEC2026), the conference brought together the vibrant openness of this coastal city with the surging innovation of supercomputing, and the deep exploration of frontier academia with the practical empowerment of the industrial ecosystem. Through a new dual-driven model of "academic flagship conference + industry expo," it wrote a new chapter of digital-real integration, sci-tech innovation empowerment, and shared prosperity in the computing industry.
Grounded in the frontiers of China's computing industry development, this year's conference focused on core hot topics such as independent innovation in domestic computing power, deep supercomputing-intelligence integration, computing interconnection, and green computing system development, while continuing and upgrading its distinctive strengths of "academic leadership, industrial integration, and ecosystem co-building." The conference gathered top talent from home and abroad, bringing together numerous academicians and experts, senior scholars, industry leaders, and elites from leading enterprises across the industrial chain, striving to create a high-level, professional HPC exchange and cooperation platform integrating frontier technology discussion, innovative achievement display, industrial resource matching, and ecosystem-wide collaboration, thereby pooling tremendous synergy for the high-quality development of China's computing industry.
The conference was hosted by the China Computer Federation (CCF), co-organized by the CCF Technical Committee on High Performance Computing and the National Supercomputing Center in Shenzhen (Shenzhen Cloud Computing Center), co-hosted by Tsinghua Shenzhen International Graduate School and the College of Computer Science and Software Engineering of Shenzhen University, and supported by the High Technology Research and Development Center of the National Natural Science Foundation of China. The conference lineup was unprecedented, gathering more than 500 experts, scholars, and industry elites in the HPC field, including experts and scholars led by more than 20 Chinese and foreign academicians and Fellows of relevant professional societies, who shared their insights. The conference featured 11 invited talks, 8 corporate presentations, 50 thematic forums, and 40 supporting special activities, and received a total of 72 Chinese paper submissions via EasyChair and 49 English paper submissions to the CCF THPC journal. The total exhibition area exceeded 12,000 square meters, bringing together 65 leading enterprises and frontier institutions, including the National Supercomputing Center in Shenzhen, the College of Computer Science and Software Engineering of Shenzhen University, Huawei, Sugon, the High-Throughput Ethernet Alliance, Xin'aisuan, Lenovo, the Jinan Supercomputing Center, Parallel Technology, Shishi Technology, and Silang Technology, with more than 5,400 industry guests registered to attend and exhibit.
At present, China's computing industry is moving from scale expansion to a new stage of high-quality development. The rapid evolution of AI large models and AI for Science continues to drive explosive growth in intelligent computing demand, and supercomputing-intelligence integration has become a confirmed industry trend. At the same time, the construction of the supercomputing internet continues to advance, and computing-power coordination, green and low-carbon development, and full-stack independent controllability of software and hardware have become core propositions for industrial development. Against this backdrop, the conference aimed to break down barriers between academic research, technological breakthroughs, and industrial application, promote the collaborative sharing of computing resources, algorithm models, and industry applications, drive the continuous iteration of domestic high-performance computing technologies, and enable computing power to better empower the digital transformation of all industries.
Intelligent Computing Tide Rises, the Greater Bay Area Builds the Foundation
To thoroughly implement the Digital China construction deployment, seize the opportunities of computing transformation in the era of large models, and promote technological innovation and industrial application in high-performance computing, intelligent computing, AI large models, and the supercomputing internet, CCF HPC2026 adopted the theme "Deep Supercomputing-Intelligence Integration, Rich Computing in Shenzhen." Grounded in the complete digital industry ecosystem of the Guangdong-Hong Kong-Macao Greater Bay Area, it conducted discussions around core topics such as the construction of a domestic computing foundation, two-way empowerment between supercomputing and large models, computing-network coordination, and green intelligent computing, helping to build an open and win-win national computing industry ecosystem and injecting strong computing momentum into the development of new quality productive forces.
On the morning of October 18, the opening ceremony of CCF HPC2026 was officially held. The opening ceremony was moderated by Researcher Tan Guangming, Vice Chair (Executive) and Office Director of the CCF Technical Committee on High Performance Computing.
Tan Guangming, Vice Chair (Executive) and Office Director of the CCF Technical Committee on High Performance Computing
Lu Yutong, Conference Chair, CCF Fellow, Professor at Sun Yat-sen University, and Director of the National Supercomputing Center in Shenzhen/Guangzhou, said in her opening remarks: "This living water of computing power is flowing to the front lines of scientific research, industrial sites, and all sectors of the economy, continuously nourishing innovation and empowering development." This year, the National Supercomputing Center in Shenzhen's "Lingsheng" fully domestic exascale supercomputing system ranked strongly at the top of the world's TOP500 supercomputers. The Guangdong-Hong Kong-Macao Greater Bay Area has two national-level supercomputing centers in Guangzhou and Shenzhen, which interact with Hong Kong and Macao's international scientific research strength and advanced manufacturing industrial system, providing unique advantages for exploring new systems, new applications, and new ecosystems of supercomputing-intelligence integration. She pointed out that China's high-performance computing must not only keep pace with the world frontier but also form its own technological system, innovation paradigm, and leadership capability in key directions, so that world-class computing power truly produces world-class scientific and engineering achievements.
Guan Haibing, Conference Chair, CCF Fellow, Chair of the CCF Technical Committee on High Performance Computing, and Vice President of Shanghai Jiao Tong University, said in his opening remarks: "Thank you very much to everyone for your support and deep participation. I also hope that in the upcoming agenda, everyone can take this opportunity to fully exchange views, deepen cooperation, and jointly inject newer momentum and vitality into the ecosystem of China's computing industry. Especially in the opening year of the 15th Five-Year Plan, at a time of unprecedented changes unseen in a century, our high-performance computing cause must plan the overall layout and get off to a good start, and science and technology must further integrate with industry to play an indispensable and important role in building a strong country in science and technology."
Professor Ni Mingxuan, Chair of the Conference and President of The Hong Kong University of Science and Technology (Guangzhou), said in his opening address: "Grounded in the Greater Bay Area, this vibrant land, the industry-academia-research collaboration in high-performance computing requires a greater strategic definition. Science knows no borders. As a hub of both domestic and international circulation, the Greater Bay Area should leverage our openness to gather innovation elements from around the world. Let us join hands to promote the deep integration of high-performance computing and artificial intelligence, and contribute more Chinese wisdom to global scientific and technological progress."
Wu Gen, Deputy Director of the High Technology Research and Development Center of the National Natural Science Foundation of China, a supporting organization of the conference, said in his opening address: "The 15th Five-Year Plan makes coordinated arrangements for computing facilities, model algorithms, and high-quality data resources, and the national integrated computing network has officially been included among the six networks to be accelerated. How to transform ever-increasing computing power into original discovery capability is an inevitable question before us scientists."
At the conference opening ceremony, Professor Chen Guoliang of the Chinese Academy of Sciences; Tan Dai, Deputy Secretary-General of the Shenzhen Municipal People's Government; Professor Liao Xiangke of the Chinese Academy of Engineering; Professor Guan Haibing of Shanghai Jiao Tong University; Researcher Zhang Yunquan, Vice Chair of the CCF Technical Committee on High Performance Computing; Li Xiaoli, Deputy Director of the Shenzhen Science and Technology Innovation Bureau; Professor Lu Yutong, Director of the National Supercomputing Center in Shenzhen/Guangzhou; Wu Gen, Deputy Director of the High Technology Research and Development Center of the National Natural Science Foundation of China; Professor Ni Mingxuan of The Hong Kong University of Science and Technology (Guangzhou); Professor Lu Kai, Vice Chair of the CCF Technical Committee on High Performance Computing; Professor Zhai Jidong, Vice Chair of the CCF Technical Committee on High Performance Computing; Professor Shi Xuanhua, Vice Chair of the CCF Technical Committee on High Performance Computing; and Mr. Zhu Zhaosheng, Executive Director of CCF, jointly inaugurated the conference.
Several major joint achievements were released at the opening ceremony. The HPC+AI Ultra-Intelligence Collaboration Open Community was launched, and the High-Throughput Ethernet (ETH+) Alliance's full-stack domestic intelligent computing network was released. A series of achievements were unveiled in a concentrated manner, covering frontier programs, open communities, domestic intelligent computing networks, and computing power applications, promoting the co-construction of an ultra-intelligence integration ecosystem and the implementation of domestic computing power innovation.
Release of the HPC+AI Ultra-Intelligence Collaboration Open Community
Release of the High-Throughput Ethernet (ETH+) Alliance Full-Stack Domestic Intelligent Computing Network
Experts Discuss the Way Forward and Chart the Future
Professor Lu Kai, Vice Chair of the CCF Technical Committee on High Performance Computing, chaired the conference report session.
Lu Kai, Vice Chair of the CCF Technical Committee on High Performance Computing
Liao Xiangke, Academician of the Chinese Academy of Engineering, computer systems software expert, and Director of Qiyuan Laboratory, said in his report "Preliminary Practice and Experience of AI4S" that AI4S, as the fifth major scientific research paradigm, is developing rapidly; the United States has launched national-level projects such as the "Genesis" program, while China faces multiple practical challenges. Qiyuan Laboratory is committed to building a "one body, three wings" architecture and establishing the "Wuzhen" foundation. He firmly believes that AI4S is not only a tool innovation but also a profound transformation of the scientific research paradigm.
In the report "Extreme Computing for Knowledge: A New Generation Ultra-Intelligence Integration System and Scientific Engineering Applications," Lu Yutong, Chair of the Conference, CCF Fellow, Professor at Sun Yat-sen University, and Director of the National Supercomputing Center in Shenzhen/Guangzhou, pointed out that using HPC+AI for Science to jointly achieve scientific research through the deep integration of numerical simulation and artificial intelligence is becoming an important new paradigm for scientific discovery. She proposed in her report: "From 'Tianhe' to 'Lingsheng,' every leapfrog development of China's supercomputing has relied on independent innovation in system architecture. We are capable of independent innovation and do not need to always follow foreign technological routes. For the next generation of supercomputing, the key lies in 'acceleration' and 'integration'—continuously improving energy efficiency with new hardware architectures, mixed precision, memory and storage, and cooling technologies, and achieving collaborative integration of multiple application modes through software and frameworks, providing a new system paradigm for highly trustworthy scientific intelligent computing in the post-exascale era."
Ding Zhaohui, Chief Technical Expert of Kunpeng High-Performance Computing, pointed out in the report "Kunpeng Gathers Strength, Ultra-Intelligence Collaboration Empowers Industrial Leapfrogging" that in the AI era, high-performance computing is evolving toward scientific research agents; AI4S multi-domain models and scientific discovery agents are rapidly being implemented, and the value of CPUs is becoming prominent in the agent era. Relying on the Kunpeng chip evolution roadmap and based on Lingqu interconnection, Huawei is building general computing super nodes, providing ultra-intelligence collaborative solutions, and implementing practices in fields such as weather forecasting, oil and gas exploration, materials simulation, and biopharmaceuticals, helping industrial digital and intelligent upgrading.
Researcher Zhang Yunquan, CCF Fellow and Vice Chair of the CCF Technical Committee on High Performance Computing, chaired the conference report session.
Zhang Yunquan, CCF Fellow and Vice Chair of the CCF Technical Committee on High Performance Computing
In the report "Sugon 8000 Empowers AI for Science Application Innovation," Cao Zhennan, Vice President of Sugon and Deputy Director of the National High Performance Computer Engineering Technology Research Center, proposed a combination of "top-level strategy + computing foundation + organized investment" to systematically reconstruct the scientific research paradigm and innovation chain. He introduced in detail the specific applications of Sugon 8000 in fields such as Earth science, materials, life sciences, astronomy, high-end equipment, and AI, and expressed full confidence in "letting computing power empower all industries."
Fu Binzhang, Co-Chair of the Technical Committee of the High-Throughput Ethernet Alliance and Head of High-Performance Networking at Alibaba Cloud, proposed in the report "Evolution of Intelligent Computing Network Architecture and Ecosystem Construction" the development path and direction of full-stack network evolution from training to inference, from HPN to TPN, and from Scale-out to Scale-up, and provided an in-depth summary of the major trends in network technology development, with the goal of achieving the integration of Scale-Out and Scale-Up and building a domestic intelligent computing network ecosystem.
In "Beyond Exascale: AI–Quantum Discovery Engines for AI for Science for Industry," Satoshi Matsuoka, Director of the RIKEN Center for Computational Science (R-CCS) in Japan and Professor at Tokyo Institute of Technology, pointed out that the goals and challenges at the current stage are "Solve intractable science problems," "Lead advanced science," "Starting from basic science," and "To societal impact."
On the afternoon of October 18, Professor Shi Xuanhua, Vice Chair of the CCF Technical Committee on High Performance Computing, chaired the keynote session.
Shi Xuanhua, Vice Chair of the CCF Technical Committee on High Performance Computing
Torsten Hoefler, Member of the Academia Europaea, Professor at ETH Zurich in Switzerland, and Director of the Scalable Parallel Computing Lab (SPCL), said in his report "Can we build an AI (Climate) Scientist?": "The scientific method is one of the greatest inventions humanity has ever made, and AI can help supercharge it."
In the report "Toward Edge-Al-Native Service Provisioning," Cao Jiannong, Vice President and Chair Professor of The Hong Kong Polytechnic University and Member of the Academia Europaea, said that "Evolving edge AI for advanced applications involves large-scale, distributed and heterogeneous resources, demanding an edge-AI-native service platform to address new challenges for scalable high performance."
Yang Zhi, co-founder of Xin'aisuan, said in the report "The Engineering Leap of Large-Model Inference and the New Industrial Landscape" that the industry has reached the eve of an inference explosion, and that industry large models will further amplify inference computing power consumption in the future. The report analyzed the situation from five dimensions: demand, efficiency, chips, form, and capital. He believes that multiple factors are jointly pushing the industry into the Token Factory era, and the inference industry will usher in a brand-new development landscape.
Xiao Limin, Chief Scientist of Lenovo Group, presented a report titled "Exploration and Practice of Storage-Compute Fusion Accelerating Large-Model Inference." In his report, he said that inference demand has surged in the Agent era, and the storage and access of KV Cache have become key requirements for large-model inference. Lenovo proposed a KV Cache storage-compute fusion acceleration system based on memory pooling and near-memory computing, which can significantly reduce inference latency and substantially improve service throughput. He believes that in the future it is necessary to continuously optimize the tiers of the storage-compute fusion pool, expand more large-model inference business scenarios, and further unleash the efficiency of inference systems in long-context scenarios.
Tao Dingwen, Research Professor and Member of the CCF Technical Committee on High Performance Computing, chaired the keynote session.
Tao Dingwen, Member of the CCF Technical Committee on High Performance Computing
Wang Xinhui, Vice President of ZTE Corporation, stated in the report "Harmonious Coexistence and Panoramic Intelligent Computing—Building an Open and Prosperous AI Ecosystem" that AI is a certain trillion-yuan track, and industry competition has shifted from single-point breakthroughs to comprehensive capability contests, with profound changes in computing power, algorithms, and data. ZTE practices the concepts of ALL in AI and AI for ALL. He proposed adhering to open decoupling, internalizing technological dividends, building a secure, controllable, harmonious and symbiotic open AI ecosystem, and promoting the inclusive development of AI and win-win industrial development.
Yang Canqun, Research Professor at National University of Defense Technology and Director of the National Supercomputing Center in Tianjin, stated in "Ultra-Large-Scale 'Thermal-Mechanical-Chemical' Multi-Effect Coupling Simulation Technology and Applications" that with the development of artificial intelligence technology, an intelligent simulation paradigm centered on data-driven and physics-constrained approaches has gradually taken shape. The GalaxyEDS simulation software supports multi-effect coupling simulation of explosion, impact, and combustion, has achieved benchmarking and surpassing in function, accuracy, and performance, and possesses domestic substitution capability.
Wang Yinglong, Director of the National Supercomputing Center in Jinan and Research Professor at Qilu University of Technology (Shandong Academy of Sciences), said in "From Trusted Computing Power to Intelligent Empowerment: Trusted Computing Power Network Technology and Applications in Digital-Intelligent Fusion Scenarios" that domestic supercomputing faces multiple practical problems such as computing power supply and demand and operations and maintenance, and that computing power services are evolving from resource supply to agent solutions. He believes that the value of computing power goes beyond computation and must be transformed into industry-oriented intelligent service capabilities.
Multi-Dimensional Collaboration, Opening Up New Prospects
On October 19-20, 50 high-quality themed forums will continue to be open, covering hot topics such as the supercomputing internet, AI large models, super-intelligence integration, industrial applications, green intelligent computing, domestic computing power ecosystems, AI for Science, and talent cultivation; the finals of the PAC Parallel Application Challenge will be fiercely contested; and the event will help realize full-industry-chain R&D and application connectivity.
Twenty-two years of diligent dedication, setting sail with vigor in the computing power era. CCF HPC2026 will continue to leverage the value of its academic and industrial aggregation platform, unite all parties across industry, academia, research, and application, continue to promote high-performance computing technology innovation, application implementation, and ecosystem co-construction, help build a solid foundation for China's independent and controllable computing power, and contribute surging computing power to the construction of a Digital China and a strong nation in science and technology. Leiphone Leiphone